Program agent harnesses with AI SDK
Introduces HarnessAgent, a unified API for running multiple agent harnesses including Claude Code, Codex, and Pi. Write your agent once and swap harnesses without rewriting code, allowing you to use the best harness available at any time.
introduces , a single API for running established agent harnesses, including Claude Code, Codex, and Pi. AI SDK has always let you switch models without rewriting your agent. Now you can switch the harness the same way.AI SDK 7HarnessAgent
Write the agent once. Use the best harness available. Today. In 3 months. A year from now.
Harnesses manage the components above a model call, including skills, sandboxes, sessions, permission flows, compaction, runtime configuration, and sub-agents. The AI SDK normalizes access to those capabilities through a unified harness abstraction.
Initial harness adapters for this experimental release include Claude Code, Codex, and Pi, with more coming soon.
Swap for or and keep the same flow. Every harness runs the agent in a sandboxed workspace, keeping the host environment safe.claudeCodecodexpiHarnessAgent
Both and return AI SDK-compatible results. If your app already uses or related AI SDK tooling, you can swap in without changing your user interface code.HarnessAgent.generate()HarnessAgent.stream()useChatHarnessAgent
is available on the AI SDK canary release. Read the to get started.HarnessAgentAI SDK harness documentation
Source: original entry ↗